Freeman et al. (2014), PNAS 111(23) — 225-study meta on active learning: exam performance +0.47 SD; odds of failing 1.95× higher under passive lecturing; robust to publication-bias checks

Claim: Freeman et al. (2014), PNAS 111(23), 8410-8415 — 225-study meta-analysis of active learning. Exam performance +0.47 SD under active learning; odds of failing 1.95× higher under passive lecturing. Robust to publication-bias checks.

Source: Freeman et al. (2014), PNAS.

Confidence: Verified (meta-analysis, top journal).

Caveat: Largest in small classes and dependent on volunteer instructors — generalisability if universally mandated is "an open question" (Freeman et al.). Education-context evidence; bridge to commercial tools is conceptual not measured.

Why this matters for Candid: The +0.47 SD is the citable single number for "active processing beats passive processing" in client conversations. Together with Chi & Wylie (2014), Educational Psychologist 49(4) — ICAP framework: Interactive > Constructive > Active > Passive engagement; ~8-10% learning improvement per step this is the strongest independent academic anchor for the engagement-via-interaction case.